Multimode Signal Distortion
In multimode operation, light waves travel down the axis of the fiber as well as a zigzag course bouncing off of the cladding. Since some of the light rays take a longer trip when they exit the far end of the core (due to its zigzag course), distortion of the original signal will occur as it recombines with the light ray that took the shorter path down the axis of the core. This results in pulse broadening at the receiver end. This distortion is called modal dispersion because the paths of the light rays are at different lengths. To counteract this multimode phenomenon, graded-index fiber was developed.
With graded-index fiber, the index of
refraction is highest along the center axis of the fiber and gradually decreases from the axis to the circumference. Light travels slower with a higher index of refraction and faster with a lower index of refraction.
With this approach, the light that travels down the center axis is deliberately slowed to match the time required for light to travel a zigzag course nearer the circumference. The result is less distortion and higher bandwidth.
Bandwidth requirements are generally not an issue with Ethernet. Multimode fibers have bandwidth specifications in frequency-distance units (MHz-km) that depend upon the operating wavelength. Doubling the distance halves the signaling rate; however, even at minimal bandwidth specifications (160 MHz-km or so), the attenuation limitations of increased fiber length will be met before the bandwidth limitations.
Lower bandwidth fiber exists with
a 200 µm core diameter. This is step-index fiber meaning that only one index of refraction exists in the core and another in the cladding. This fiber is intended for shorter runs and is easier to connect and is more resilient to physical abuse due to its larger core size. This fiber is found in plant floor applications but is not recommended for Ethernet.
|
|
|
Figure 3 — Total internal reflection occurs when the incident angle
exceeds the critical angle.
|
Operating Wavelengths
Fiber optic transmitters and receivers are generally classified to operate in either of three frequencies. These frequencies have been found to have the lowest attenuation across a band of frequencies. The regions of lowest attenuation are called windows. The particular frequencies
the industry uses are 850 nm, 1300 nm and 1550 nm. The two lower wavelengths offer cost/performance tradeoffs that are of interest in Ethernet applications. The 850 nm technology is readily available at the lowest cost. However, fiber optic cable attenuation is higher in the 850 nm band than in the 1300 nm band and the bandwidth is less. This attenuation is what limits the fiber optic segment lengths when using Ethernet. The 1300 nm receivers and transmitters are more costly but are recommended when long distances are to be encountered or 100 Mbps operation is required. The 850 nm
technology is generally used with multimode applications, while the 1300 nm technology is used with either single-mode or multimode operation. Because of cost, the 1550 nm technology is not popular with Ethernet.
Fiber Optic Transmitters
Both 850 nm and 1300 nm fiber optic
transmitters can be found in hubs and network interface modules (NIMs) and the two technologies cannot be mixed. These transmitters are available with either ST, SC or MIC connectors. The ST connector operates similar to a small coaxial BNC connector. It prevents over-tightening and provides repeatable insertion loss. The SC connector is
a low-cost, snap-in connector while the similar style MIC connector was originally intended for Fiber Distributed Data Interface (FDDI) applications.
In fiber optic implementations, a
separate transmitter and receiver are used instead of a transceiver. Fiber optic links use a duplex cable for NIM-to-hub and hub-to-hub connections. A transmitter at point A connects to a receiver at point B. Point B’s transmitter attaches to point A’s receiver. Therefore, a crossover function must be accomplished in the cabling. Transmitters and receivers may or may not be color-coded so care must be exercised to pair a transmitter to a receiver.
|
|
|
Figure 4 — Step-index fiber has the lowest bandwidth,
while single-mode fiber offers the highest.
|
Transmitter Power
Transmitters are rated in dBm with
0 dBm corresponding to 1 milliwatt of power. Transmitter output can vary from device to device, so it is important to 100% test transmitters to ensure that none are shipped below the minimum specified in the Ethernet standard. Testing is usually accomplished by applying a square wave signal and measuring the average power with an optical power meter. Transmitter output also depends upon the fiber size. More energy is launched into larger fiber sizes; therefore, a power rating shown in a specification is based upon a particular core size.
Receiver Sensitivity
Receiver sensitivity is also rated in dBm and is based upon receiving the same square wave signal generated by the transmitter. Typically only a maximum sensitivity rating is given which represents the weakest signal discernable by the receiving electronics. Separate receivers are required for 850 nm and 1300 nm operation. Receiver sensitivity is typically the same over a batch of receivers and does not exhibit the same variability as transmitters.
Optical Power Budget
When specifying a fiber
optic installation, attention must be paid to the available optical power budget. The power budget is the difference between the light source strength minus receiver sensitivity expressed in dB. This value must be compared to the link loss that is the total attenuation due to optical cable and optical connectors. The link loss
must be less than the power budget. The difference is called the power margin that provides an indication of system robustness.
Link Loss
To determine the link loss, all losses due to fiber length and cable connections must be summed. Fiber optic cable attenuation is usually specified by the cable manufacturer. Use this figure to determine the attenuation for a particular length of fiber cable. It is also necessary to include losses due to cable terminations. Connectors usually create a loss of from 0.5 to 1 dB for each connection. For example, assume a 1500-meter run of 62.5 µm cable which the cable manufacturer specifies as having a cable attenuation of 3.5 dB per 1000 meters. The cable loss would therefore be 5.25 dB. Assume there are two connector losses of 0.5 dB each for a total of 1 dB. The link loss would therefore total 6.25 dB. If the light
source produced –20 dBm and the receiver sensitivity is –30.4 dBm, then the power budget would be 10.4 dB which is greater than the link loss by 4.15 dB. This difference would represent a high degree of margin since a 3 dB margin is what is typically recommended to account for aging. Recommendations on acceptable attenuation values can be found in TIA/EIA-568-A Commercial Building Telecommunications Cabling Standard.
Overdrive
Overdrive occurs when too little fiber optic cable is used resulting in insufficient attenuation; thereby, saturating the receiver. To correct this condition, a longer length of fiber optic cable must be installed between the transmitter and receiver. This is potentially a problem
with larger core cable. Another solution is to have a receiver with a wide dynamic range. Over this range, the receiver will accept varying levels of signal without overload.
Delay Budget
People frequently assume that with fiber optics, signals propagate at the speed of light. This is not true. In fact, the propagation factor is 0.67c or 5 ns/m which is slower than an electrical signal over coaxial cable. The delay through cables and hubs is an issue for shared Ethernet systems that operate over half-duplex links and must obey the rules for collision detection. It is not an issue for full-duplex links which avoid collisions altogether.
|
